agentleFS
iwe-orgiwe-org/dev-workspace/.claude/skills/setup/SKILL.md

setup

Skill11 starsChanged 2 days agoMITView on GitHub

Brownfield-first workspace onboarding — scan the codebase and draft data/product.md plus starting architecture and spec docs from what the code shows, interview the developer only for what the code can't answer, mark the onboarding tasks done, and remove the example documents and everything they left behind. Use when the user says "run the setup", "set up the workspace", or when the "What is it" section of data/product.md is still an unfilled ✏️.

191 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use add-voice-transcription, add-discord and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • juancruzmunozalbeloSkill

    add-voice-transcription

    Add voice message transcription to NanoClaw using OpenAI's Whisper API. Automatically transcribes WhatsApp voice notes so the agent can read and respond to them.

    Also in 5 projects that use this

  • MicrosoftSkill

    add-discord

    Add Discord bot channel integration via Chat SDK.

    Also in 11 projects that use this

  • MicrosoftSkill

    add-slack

    Add Slack channel integration via Chat SDK.

    Also in 11 projects that use this

  • juancruzmunozalbeloSkill

    add-telegram-swarm

    Add Agent Swarm (Teams) support to Telegram. Each subagent gets its own bot identity in the group. Requires Telegram channel to be set up first (use /add-telegram). Triggers on "agent swarm", "agent teams telegram", "telegram swarm", "bot pool".

    Also in 7 projects that use this

  • +8 more from microsoft

The file

We split this skill into the rules it gives an agent

9 rules. 0 are common, 5 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 168 edited versions we read keep these.

  • 5

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Assess: Read data/product.md and list which ✏️ blocks are unfilled.

    line 14
  2. Distinctive:

    Locate the codebase, and settle git: Ask where the project's code lives (often the parent or a sibling directory of this workspace, or this …

    line 17
  3. Distinctive:

    Scan: Read the repository's README, package manifests, build configuration, entry points, directory layout, and test setup.

    line 24
  4. Distinctive:

    Interview: Ask in batches, conversationally:

    line 35
  5. Very specific:

    Remove the examples: Delete every *-example.md and releases/example-*.md under data/ with iwe delete, which also removes their inclusion …

    line 56
  6. Very specific:

    Stamp and validate: Every document this run changed gets a generated stamp (AGENTS.md, step 5) — the hubs and the onboarding tasks …

    line 86
  7. Distinctive:

    Never invent — every drafted statement must trace to the code or the developer's answer.

    line 94
  8. Very specific:

    Never leave a ✏️ block half-filled — ask follow-ups until each section is concrete.

    line 96
  9. Distinctive:

    Don't delete the examples until real documents exist for the developer to learn the conventions from (product.md filled is the minimum).

    line 103

“Common” is measured: kept by at least 6 in 10 of the 168 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Workspace setup

Turn the blank workspace into this project's memory. The codebase is the primary source — read it first and draft from evidence; ask the developer only what the code can't answer. Every future session reads what you write here.

Steps

  1. Assess. Read data/product.md and list which ✏️ blocks are unfilled. If ## What is it is filled, setup has already run: ask what to revise, or offer to fill the sections that were skipped, instead of onboarding again.
  2. Locate the codebase, and settle git. Ask where the project's code lives (often the parent or a sibling directory of this workspace, or this workspace sits inside the repository itself). Then check whether the workspace is already inside a git repository (git rev-parse --is-inside-work-tree). If it is not, ask once whether it should be one, and run git init only on a yes. A no is final: the skills validate and skip the commit. If the project is greenfield — no code yet — skip to step 4 and run the interview alone.
  3. Scan. Read the repository's README, package manifests, build configuration, entry points, directory layout, and test setup. Draft from what you find:
    • data/product.md — What is it, Platforms, Stack from direct evidence; leave ✏️ plus a concrete question under any section the code can't answer (Users, Constraints usually need the developer).
    • One starting data/architecture/<slug>.md describing the module layout and any design decisions visible in the code (state management, storage, process boundaries). Say "unknown" where you'd be guessing.
    • Propose (don't yet write) spec docs for the 2–3 most load-bearing behaviors you can identify.
  4. Interview. Ask in batches, conversationally:
    • Product: the one-liner; who uses it and for what; who it's not for.
    • Reality: current stage, what's shipped vs. aspirational, the next thing they intend to build.
    • Constraints: performance budgets, compatibility promises, licensing, privacy — anything every plan must respect.
    • Rules: recurring instructions they find themselves repeating to agents or contributors — these become the Authoring rules section. It is optional: if there are none, replace its ✏️ with "None yet."
  5. Write. Fill the data/product.md sections, deleting the italic instruction lines as sections fill, and replace the bullet under ## Changelog with - <today> — product doc filled in by setup. Write the confirmed architecture doc(s) and any spec stubs the developer approved, linking each from data/architecture.md / data/spec.md.
  6. Close the onboarding tasks. The four onboarding tasks in data/backlog/ carry the template's date; give them today's: iwe update -k data/backlog/fill-product-doc -k data/backlog/capture-current-architecture -k data/backlog/import-existing-backlog -k data/backlog/plan-first-feature --set created=<today>. Then mark the finished ones — fill-product-doc once no ✏️ is left in data/product.md, and capture-current-architecture if step 3 ran: iwe update -k data/backlog/<slug> --set stage=done --set completed=<today>, and move their links in data/backlog.md to ## Done.
  7. Remove the examples. Delete every *-example.md and releases/example-*.md under data/ with iwe delete, which also removes their inclusion links from the hubs:

    for f in data/*/*-example.md data/releases/example-*.md; do iwe delete "${f%.md}"; done

    Then remove what the examples left behind. Each of these is a separate edit; do all of them:

    • data/log.md — replace every dated group with a single one, ## <today>, holding one bullet: - **Initialization**: Workspace set up — [product](product.md) filled in and the example documents removed. The file needs at least one dated group to stay valid.
    • data/plans-archive.md — delete the ### YYYY-MM headings under ## Done; they are the months the example plans shipped in.
    • data/releases/metrics.csv — cut it down to its header line; its rows describe the example releases.
    • AGENTS.md — delete the Example docs paragraph under Conventions.
    • SCHEMA.md — delete the paragraph that begins "The examples below describe one fictional product".
    • Anywhere else: iwe delete turns an inline link to a deleted document into plain text, so a document you wrote during setup that cited an example now has a sentence without its link; reword or remove it. No link to an example may be left: grep -rn -e '-example' -e 'example-' data index.md must print nothing.

    The hub charts draw empty axes until real documents arrive, and data/releases/unreleased.md stays, empty, as the page the first shipped work is linked from.

  8. Stamp and validate. Every document this run changed gets a generated stamp (AGENTS.md, step 5) — the hubs and the onboarding tasks included, which still carry the template's. Then iwe normalize, and iwe schema validate — both must pass clean. If the workspace is a git repository, commit with a message like setup: product doc filled, architecture captured, examples removed.

Rules

  • Never invent — every drafted statement must trace to the code or the developer's answer; "unknown" beats plausible fiction.
  • Never leave a ✏️ block half-filled — ask follow-ups until each section is concrete. ## What is it has to be filled for setup to finish. The developer may say "skip for now" to any other section: leave its ✏️ so the gap stays visible, and leave fill-product-doc planned. Later sessions mention the gap once and do not run setup again.
  • Propose before bulk-writing: spec stubs and extra architecture docs get a one-line pitch each and the developer's yes before they exist.
  • Don't delete the examples until real documents exist for the developer to learn the conventions from (product.md filled is the minimum).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.